Research On AOI Oriented Structured Light Imaging System

As the core component of electronic manufacturing industry,PCB can only be put into the market after strict defect detection during tin loading,assembly,welding and other production processes.Automatic optical inspection(AOI)is a non-contact detection technology for product defects based on the optical imaging principle.The3 D AOI system designed based on the principle of structured light imaging can obtain the rich 3D surface information of the object to be tested,which is of great significance for the increasingly precise and automatic PCB quality control.After introducing the principle of structured light imaging in detail,this paper conducted in-depth research on Schempflug structure,projector calibration,multi-eye coupling calibration and other issues,and proposed a high-precision,high-coupling multi-eye structured light system calibration scheme Finally,the high-precision geometric parameters obtained by the calibration are introduced into the imaging algorithm to obtain the three-dimensional information of the target surface to be measured,and a real-time optimization scheme is further proposed for noise and other problems.Finally,high-precision and high-quality point cloud data is obtained.The full text of the study is as follows:(1)A set of 3D based on the principle of structured light imaging is designed for the application of circuit board defect detection AOI system,considering the occlusion,shadow and other problems caused by the limitation of view angle,adopts an imaging structure composed of one projector and four side view cameras,and four side view cameras are equipped with tilt lens meeting the condition of schempflug to increase the common depth of field of structured light imaging.(2)A complete set of high-precision calibration scheme is proposed.Firstly,a tilt model characterized by two-dimensional tilt angle is used to solve the description problem of single camera pinhole model with the introduction of schempflug structure,and then the problem of control point ext raction of projector is solved by using the method of mark reconstruction based on phase mapping,and it is optimized by using multiple perspectives.Finally,aiming at the problem of multi view calibration,a multi-objective calibration method based on master-slave coupling is proposed,and a global calibration objective function is established for the system by using the idea of beam adjustment method.The calibration parameters of the system are obtained through optimization,which can avoid complex poin t cloud registration operation and save a lot of calculation time.(3)Based on the system geometry information obtained from high-precision system calibration,combined with structured light encoding and decoding algorithm and geometric imaging algorithm,the high-precision point cloud data to be measured is generated.Aiming at the real-time processing of the generated point cloud noise,a point cloud quality optimization scheme based on the edge window filter of the depth map is proposed,it makes full use of the "physical edge preserving" feature of the edge window filter and its flexible operation feature.At the same time,it realizes the denoising and smoothing of the point cloud,and the processing speed is about 60 times faster than the traditional three-dimensional filtering method.The structured light system designed in this paper aims at the scene of circuit board defect detection.Through the point cloud measurement experiment of the calibrated control points,it is proved that the point cloud m easurement error is less than 6um,which can meet the demand of high precision measurement.